KMC stay silent when output file is not created due to memory problem

[cmonat]

FYI: KMC stays silent when the output file is not created due to a memory problem.
In my case, I'm running KMC3 on a SLURM cluster, and my log files (.err and .out) are normal (no errors in the .err and the standard KMC information in the .out); the only problem is that the .kmc_sup and .kmc_pre files are not created.
My colleague encountered the same problem on an AWS cluster.
Increasing the -m parameter solves the problem, but it would be nice if KMC informed the user that the output files are not created due to a memory problem.

[marekkokot]

Thanks, this is weird. I would like to know the cause of this memory issue. I mean it should not be required to increase -m. If KMC needs more memory for the given dataset, it will just allocate it (unless -sm is also provided). I would like to reproduce this issue. You mentioned that you cannot share your input data on which this error occurs, but perhaps you also experience this error with some other public data? I would like to try this on non-Slurm, and if the problem is only on slurm, it would be great to have the slurm config.